Fibroblast growth factor receptor inhibitors for non-muscle-invasive bladder cancer: THOR-2 and beyond

High-risk non-muscle-invasive bladder cancer (NMIBC), which encompasses carcinoma in situ (CIS), high-grade T1, and selected high-grade Ta tumors (1), is typically managed by surgical resection and Bacillus Calmette-Guérin (BCG) instillations as frontline treatment, but over half of patients develop recurrence (2). Many patients with recurrence after BCG can proceed with cystectomy, but bladder-sparing intravesical chemotherapy options, including chemotherapy, can also be utilized (3). For patients with CIS recurrence, novel intravesical therapies such as nadofaragene firadenovec, nogapendekin alfa inbakicept in combination with BCG, and the gemcitabine intravesical system TAR-200 have all been recently approved by the United States Food and Drug Administration (4-6), but there are no regulatory approvals for patients with recurrent Ta/T1 tumors without CIS component. Therefore, management of high-risk BCG-exposed or refractory Ta/T1 tumors without CIS is an unmet need, as patients with NMIBC recurrence have an increased risk of having disease progression to muscle-invasive disease. Increasingly, with a deeper understanding of bladder cancer biology, additional biomarker-driven approaches are being explored. This commentary focuses on the results of THOR-2 (NCT04172675), which investigated the efficacy of the oral fibroblast growth factor receptor (FGFR) inhibitor erdafitinib in patients with high-risk and intermediate-risk NMIBC with FGFR3/2 alterations, including one cohort dedicated to Ta/T1 tumors without CIS component.

FGFR is a driver of bladder cancer development, inducing cell proliferation and survival (7). Around 60–80% of patients with NMIBC may harbor FGFR alterations, most commonly activating point mutations and gene fusions (8). NMIBC with FGFR alterations have been noted to have a more favorable prognosis (9), and further FGFR alterations are associated with the luminal subtype of bladder cancer (10). In the muscle-invasive and metastatic setting, the rate of FGFR3 alterations specifically is lower, at around 15–20%. This lower rate is explained by advanced disease being represented by tumors that are more heterogeneous with more aggressive biological features and other genetic alterations (7). The success of FGFR3-targeted therapy in the metastatic setting provides rationale for evaluating these agents in the NMIBC setting to prevent or delay recurrence and disease progression.

Erdafitinib is a pan-FGFR 1, 2, 3, and 4 tyrosine kinase inhibitor. In the BLC2001 (NCT02365597) phase 2 trial of patients with treatment-refractory metastatic urothelial carcinoma (mUC), the objective response rate was 40%, the median progression-free survival (PFS) was 5.5 months, and the median overall survival (OS) was 13.8 months (11). THOR cohort 1 (NCT03390504) was a confirmatory phase 3 trial (12) for patients with mUC treated with 1–2 prior lines of therapy, including an immune checkpoint inhibitor, who were randomized 1:1 to receive oral erdafitinib 8 mg daily versus either docetaxel or vinflunine. A total of 266 patients were randomized in this cohort, and the median OS was 12.1 months for erdafitinib versus 7.8 months for chemotherapy (HR 0.64; 95% CI: 0.47–0.88, P=0.005), confirming the role of erdafitinib in this population, although more than 10% of patients had to discontinue erdafitinib early due to toxicity.

Given the OS benefit of erdafitinib for patients with mUC with FGFR alterations and predominance of FGFR alterations amongst patients with NMIBC, THOR-2 evaluated the efficacy of oral erdafitinib for patients with high-risk or intermediate-risk NMIBC with FGFR3/2 alterations. Among alterations, the included FGFR3/2 mutations were R248C, S249C, G370C, Y373C, and FGFR2 or FGFR3 gene fusions included FGFR2-BICC1, FGFR2-CASP7, FGFR3-TACC3, and FGFR3-BAIAP2L1. To assess FGFR status, all biopsy samples were sent to a central laboratory for testing. Eligible patients had Eastern Cooperative Oncology Group (ECOG) performance status 0–1, no significant organ dysfunction, and refused or were ineligible for radical cystectomy. The trial included three cohorts with primary efficacy endpoint based on cohort 1. This cohort consisted of patients with high-risk NMIBC papillary-only (Ta/T1) disease including both BCG-experienced and BCG-unresponsive patients, and randomization was 2:1 to either oral erdafitinib or intravesical chemotherapy. Intravesical chemotherapy consisted of either mitomycin C/hyperthermic mitomycin C at 40 mg or gemcitabine 2,000 mg weekly for four or more induction doses followed by monthly maintenance for at least 6 months, with more doses allowed per local standard of care. Erdafitinib was planned for a maximum of 2 years unless there was high-risk disease recurrence, toxicity, or patient/investigator decision.

The primary endpoint of the trial was recurrence-free survival (RFS) in cohort 1, defined as time from randomization to the first documented recurrence, progression to muscle-invasive disease, or death from any cause. Cohort 2 included patients with BCG-unresponsive high-risk NMIBC CIS with FGFR3/2 alterations with or without papillary tumor. Cohort 3 included patients with recurrent low-grade intermediate risk-NMIBC, with <5% risk of progression at 2 years, and >50% risk of recurrence using the European Organization for Research and Treatment of Cancer risk calculator. The exploratory endpoints for cohorts 2 and 3 were complete response (CR) rates and duration of response. CR in cohort 2 was defined as having a negative cystoscopy and negative/atypical urine cytology, or a cystoscopy with biopsy-proven benign findings or low-grade NMIBC and negative cytology. CR in cohort 3 was defined as complete disappearance of the lesion, with no new lesion or viable tumor seen on histopathological examination, and negative urine cytology. The median follow-up was about 1.5 years for all cohorts at the time of reporting.

THOR-2 encountered several difficulties and was terminated to enrollment early in December 2022 due to slow accrual. The impact of the coronavirus disease 2019 (COVID-19) pandemic and the intermittent shortage of BCG causing a lack of eligible patients contributed to the early termination of the study. Concerns regarding systemic toxicity of oral erdafitinib for patients with localized cancer by treating oncologists and urologists also hindered enrollment. Notably, significant resources were required to screen patients for FGFR alterations. Of 1,714 patients screened, 1,258 patients had adequate tumor samples for testing, and 431 patients (34.2%) had FGFR3/2 alterations, which was a lower rate than expected. Factors accounting for this may have included that BCG-unresponsive high-risk NMIBC may have a lower rate of FGFR3/2 alterations, patients having potentially less sensitive polymerase chain reaction testing instead of next-generation sequencing (either was allowed), and tumor heterogeneity.

Cohort 1’s target enrollment was 240 patients, but only 73 patients were accrued, with 49 patients randomized to erdafitinib and 24 patients to intravesical chemotherapy. The patient demographics appeared well balanced between these two groups. Median age was 68–69 years, approximately 76–79% of patients were male, 50–55% were white, and 58–59% were tumor stage Ta compared to 41–42% tumor stage T1. Patients received erdafitinib for a median duration of 12.8 months [interquartile range (IQR) 5.4–20.3 months] versus intravesical chemotherapy for a median duration of 6.2 months (IQR 5.9–14.4 months). The median RFS was not reached for erdafitinib versus 11.6 months for intravesical chemotherapy (HR 0.28; 95% CI: 0.13–0.61; P=0.0007). The benefit of erdafitinib was seen regardless of tumor stage, with Ta (HR 0.31, 95% CI: 0.11–0.85) and T1 (HR 0.22, 95% CI: 0.07–0.76); or prior BCG therapy, with BCG-experienced HR 0.13 (95% CI: 0.03–0.67) and BCG-unresponsive HR 0.34 (95% CI: 0.13–0.89). All 49 patients who received erdafitinib had at least one erdafitinib-related grade ≥1 adverse event, 16 patients (33%) reported grade ≥3 treatment-related adverse events (TRAEs), and 6 patients (12%) had serious erdafitinib-related adverse events. Thirty-six patients (73%) had erdafitinib-related adverse events leading to dose reduction, and 37 patients (76%) had erdafitinib-related adverse events leading to drug interruption. Fifteen patients (31%) had to discontinue erdafitinib, including 14 patients (29%) where discontinuation was deemed related to erdafitinib. In contrast, out of 23 patients who received intravesical chemotherapy, only 9 patients (39%) had at least one grade ≥1 TRAE, including 1 patient (4%) with grade ≥3 TRAE, and no serious chemotherapy-related adverse events. No patients needed to discontinue intravesical chemotherapy due to adverse events, and only 2 patients (9%) had TRAEs leading to chemotherapy interruption.

For cohorts 2 and 3, 16 and 18 patients were enrolled, respectively. In cohort 2, the CR rate at 32 weeks was 81% (95% CI: 54–96%) with a median duration of response of 23.3 months. Five patients had recurrence or progression after initial response. In cohort 3, the CR rate was 89% (95% CI: 65–99%) at 3 months, and the median duration of response was not reached. The toxicity profiles in these cohorts were similar to cohort 1. In cohort 2, grade ≥3 TRAEs were reported in 56% of patients, and erdafitinib-related adverse events led to dose reduction in 63%, erdafitinib interruption in 69%, and erdafitinib discontinuation in 13% of patients. For cohort 3, these values were 17%, 61%, 72%, and 22%, respectively.

The results do suggest that oral erdafitinib was more promising than intravesical chemotherapy based on a statistically significant RFS improvement in cohort 1, but this remains hypothesis-generating due to the small sample size from early termination. Nevertheless, it remains an open question whether RFS improvement correlates with other important outcomes in NMIBC, including high-grade RFS, progression to muscle-invasive or metastatic disease, avoidance of radical cystectomy, bladder preservation, and preservation of health-related quality of life. These details were not published and are unlikely to be fully addressed with THOR-2. In particular, health-related quality of life is essential in evaluating a medication like erdafitinib with potentially life-altering toxic adverse effects.

Importantly, there were significant systemic toxicities in all cohorts treated with oral erdafitinib which required medical intervention. This happened despite the lower starting dose of erdafitinib at 6 mg daily. This dose reduction occurred after the original four patients of THOR-2 received 8 mg, which is commonly used in the metastatic setting, before the independent data monitoring committee recommended lowering the starting dose to 6 mg daily to improve tolerability while maintaining activity. The toxicities were consistent with the expected side effects of erdafitinib, including nail, skin, and ocular toxicities, hyperphosphatemia, and stomatitis, which all occurred in more than half of patients. Hyperphosphatemia was the most common TRAE, occurring in 63 of the 83 patients (76%) who received erdafitinib. Central serous retinopathy events were reported in 27 patients (33%), and 10 patients (12%) still had unresolved retinopathy at time of data cutoff. Unfortunately, patient-reported outcome measures were not reported in THOR-2, which could have provided valuable information on whether patients’ health-related quality of life measures truly worsened while receiving erdafitinib.

These toxicities are all associated with the FGFR inhibitor class effect. FGFR inhibition interferes with FGF23 signaling controlled by FGFR1 and FGFR4, leading to increased reabsorption of phosphate from proximal tubules of the nephron and resulting in hyperphosphatemia (13). FGFRs are important in the homeostasis, proliferation, and differentiation of epidermal tissue, and pan-FGFR inhibition disrupts these processes (14). FGFRs are also active in the retina and ocular surface tissues (15). Among all 83 patients in THOR-2 who received erdafitinib, 22 patients (27%) had to discontinue erdafitinib. This is much higher than for patients with mUC in THOR, in which 14.1% of patients with metastatic disease discontinued erdafitinib due to adverse events, with 8.1% discontinuing due to erdafitinib-related adverse events. While erdafitinib may improve patient outcomes in the NMIBC population, the benefit-to-risk ratio is considered quite differently in this population relative to metastatic patients, and there is less willingness on the part of both patients and physicians to tolerate this degree of toxicity. Therefore, there is a much lower threshold for patients with localized disease without any distant spread to discontinue a systemic drug compared to patients with metastatic disease.

Systemic therapies utilized in the metastatic setting have previously been explored in BCG-experienced NMIBC. Pembrolizumab received regulatory approval for patients with BCG-unresponsive high-risk NMIBC with CIS in January 2020 based on the results of KEYNOTE-057 (16). In this single-arm trial, the 12-month CR rate was 19% for cohort A (patients with NMIBC CIS). However, grade ≥3 TRAEs were reported in 13% of patients, and overall, about 20% of patients developed immune-related adverse events. Therefore, pembrolizumab has not been widely adopted for patients meeting this indication, highlighting the difficulties of using systemic therapy for a non-metastatic cancer with high rates of OS, where the short-term and risks of long-term toxicities may outweigh benefits. Oral erdafitinib would likely encounter similar challenges in terms of potential utilization for FGFR-altered NMIBC.

In the past five years, three intravesical agents have achieved regulatory approval for patients with high-risk BCG-unresponsive NMIBC CIS with or without papillary tumors: nadofaragene firadenovec, nogapendekin alfa inbakicept in combination with BCG, and TAR-200. Because these agents are directly instilled into the bladder, they have lower risks of systemic toxicities relative to systemic therapies. Although cross-trial comparisons should be avoided, the lower rates of adverse events with these intravesical therapies contrast with what was observed with oral erdafitinib (4-6). However, unlike THOR-2, these trials did not utilize a biomarker-based selection process for patients. Additionally, all these trials were single-arm studies, while THOR-2 was the only trial to include a randomized treatment arm compared to a control arm of intravesical chemotherapy. In cohort 1 of THOR-2, intravesical mitomycin or gemcitabine was a reasonable, although more outdated, option for patients who either declined or were ineligible for radical cystectomy (17). The intravesical sequential chemotherapy combination of gemcitabine and docetaxel has become more widely adopted for patients with BCG-unresponsive NMIBC (18). With the rapidly changing landscape of NMIBC including these novel intravesical therapies, the design of future trials requires careful selection of an up-to-date control arm that would not exaggerate potential benefits of the experimental arm.

Given the data and lessons learned from THOR-2, it is unlikely for oral erdafitinib to be widely adopted for use in patients with NMIBC given the extensive toxicity profile. A systemic drug must be very selective for a specific target to minimize toxicities, and next generation of more selective FGFR3 inhibitors may represent more suitable options in this setting. Potential examples include dabogratinib and vepugratinib, which are selective FGFR3 inhibitors which spare other FGFR targets and thus could hopefully minimize associated toxicity. Dabogratinib is being explored in SURF302 (NCT06995677) for patients with low-grade intermediate-risk FGFR3-altered NMIBC, with estimated completion in 2028 (19). Vepugratinib was initially evaluated in FORAGER-1 for patients in the metastatic setting (20). The treatment regimen of 200 mg twice daily only required dose reductions in 14% of patients and discontinuation in 3% of patients due to treatment-emergent adverse events while suggesting promising efficacy. Vepugratinib’s efficacy and safety will be further explored in the FORAGER-2 (NCT07218380) phase 3 trial in the metastatic setting (21).

Alternatively, there is opportunity to target FGFR alterations with intravesical therapy. One option is TAR-210, an investigational intravesical drug delivery system to continuously release erdafitinib in the bladder. It is anticipated that with intravesical therapy, systemic toxicities of erdafitinib can be minimized. TAR-210 is being evaluated in the MoonRISe clinical trials program to treat NMIBC with FGFR alterations. MoonRISe-1 (NCT06319820) compares TAR-210 for one year against intravesical chemotherapy in patients with FGFR-altered low-grade intermediate-risk NMIBC, and MoonRISe-3 (NCT06919965) compares TAR-210 for two years versus intravesical chemotherapy for patients with FGFR-altered BCG-treated high-risk NMIBC (22,23). Both of these are phase 3 trials that are expected to finish in 2032.

In summary, the investigators of THOR-2 and authors of the manuscript should be commended for addressing this important question using a biomarker-driven randomized trial incorporating an intravesical chemotherapy control arm. This was an ambitious effort that required a substantial amount of time and resources, highlighting the challenges of precision therapy in NMIBC. Although erdafitinib demonstrated clinical activity, its unfavorable toxicity profile limits its applicability in a disease with favorable long-term survival. Agents with more specific targeting of FGFR3 and intravesical targeted therapies to deliver drugs with reduced systemic exposure, may offer more potential. However, biomarker-driven trials for NMIBC may continue to face enrollment challenges given the competitive landscape of intravesical therapies that do not require molecular selection. THOR-2 therefore is a practice-informing trial, providing important insights to guide future trials and drug development in NMIBC.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.